- Be able to discuss the main need and the main reasons for the transition from IPv4 to IPv6.
- Understand that the Internet is a virtual network.
- Understand how IP datagrams are encapsulated for transmission across physical networks.
- Understand why IP addresses are necessary and how the use of IP addresses facilitates the transmission of IP datagrams across the Internet.
- Understand the formats of IPv4 class-full and classless addresses, as well as the format of IPv6 addresses. This includes understanding the significance of network prefixes and host suffixes, as well as address masks.
- Understand dotted decimal notation
- Be familiar with the following special types of IPv4 addresses:
- Network Address
- Directed Broadcast Address
- Limited Broadcast Address
- "This Computer" Address
- Loopback Address
- Understand that IP and other host addresses are assigned to an interface of a host to a network, not really to the host itself.
- Understand what multihomed means.
- Understand IPv6 colon-hex address notation.
- Beginning of description of Internet protocol software
- Addressing scheme (IPv4)
- Address masks
- IPv4 and IPv6 addressing
- Classless and subnet addressing
- IPv4 uses 32-bit addresses.
- The original design for the Internet was published in 1974.
[Vinton G. Cerf, Robert E. Kahn, "A Protocol for Packet Network Intercommunication", IEEE Transactions on Communications, Vol. 22, No. 5, May 1974 pp. 637-648] - 32-bit addresses were already in use in 1974, notably in the ARPAnet, which was the precursor of the Internet.
- In 1974, network designers did not anticipate the explosive growth that would occur in the size of the Internet.
- Gradually it became clear that the 32-bits would be exhausted completely, and that a new addressing scheme would have to be adopted to allow for continued growth.
- The address space limitation was the primary motivation for the development of IPv6, version 6 of the Internet Protocol.
- IPv4 was central to the workings of the Internet, and there was reluctance to transition to IPv6.
- People invented ways to help the 32-bit address scheme to remain viable longer, and to delay adoption of IPv6.
- The Internet is a virtual network existing at a level of abstraction above the physical networks that support it.
- To maintain the illusion of the Internet - a single virtual network that connects computers on a multiplicity of real physical networks, the designers of the Internet chose a uniform addressing scheme, packet format, and delivery technique.
- The design of Internet addresses and their assignment to NIC's is independent of the addressing scheme used on the physical networks.
- Because different types of MAC addresses are incompatible, and because the Internet requires hierarchical addressing, it would be impossible to use MAC addresses as Internet addresses.
- Packets used for Internet communication are generally quite similar to typical MAC frames. Internet packets contain the (IP) addresses of the source and destination, a payload, and a check sum.
- An Internet Protocol address (aka IP address, aka Internet address) is a 32-bit (IPv4) or 128-bit (IPv6) number.
- Packets traveling across the Internet are supposed to contain two IP addresses: that of the sender (the source address) and that of the intended recipient (the destination address).
- Example: The IP address of cs.csustan.edu is 130.17.70.80. Each of the parts of the address stands for an 8-bit number ( 4*(8 bits)=32 bits )
- IP addresses are hierarchical. They have two parts: a prefix and a suffix.
- A prefix is assigned by an authority to one, and only, one network.
- All hosts on a given network have the same prefix.
- Example: the prefix for the csustan.edu network is 130.17. All computers on our campus network have IP numbers that start with 130.17.
- It is very important to NOT assign the same IP address to two different NIC's. Since the prefixes of two computers on the same network have to be the same, that means the suffixes have to be different.
- Example: cs.csustan.edu: 130.17.70.80
- Example: www.cs.csustan.edu: 130.17.70.35
- In the original IPv4 addressing scheme, there were five different
classes of address:
- Class A: first bit is 0, prefix is 8 bits, suffix is 24 bits - provides addresses for about 128 big networks with many hosts.
- Class B: first two bits are 10, prefix is 16 bits, suffix is 16 bits - provides addresses for about 16,000 medium-sized networks with up to about 64,000 hosts
- Class C: first three bits are 110, prefix is 24 bits, suffix is 8 bits - provides addresses for about two million small networks with up to about 256 hosts.
- Class D: first four bits are 1110 - used for multicast addresses
- Class E: first four bits are 1111 - reserved (not assigned)
- Classes A through D are in use, although Class D addresses (for multicast) are used only locally, within individual networks.
- Example: castor.csustan.edu has IP address 130.17.70.83
130 = 128+2 = 1000 0010 17 = 16+1 = 0001 0001 70 = 64+4+2 = 0100 0110 83 = 64+16+2+1 = 0101 0011
So the binary equivalent of 130.17.70.83 is the 32-bit number:1000 0010 0001 0001 0100 0110 0101 0011
- Note that, since they represent unsigned 8-bit binary numbers, each of the four sections of a dotted decimal is in the range 0 to 255. It can't be any larger than 255 (or less than 0).
- Notice that an octet whose bits are all 1-bits (1111 1111) has the dotted decimal value of 255.
- A company or organization usually gets its network prefix(es) assigned to it by an ISP.
- There are registrars who make blocks of addresses available to ISPs.
- The Internet Corporation for Assigned Names and Numbers (ICANN) authorizes the registrars.
- ICANN has the ultimate responsibility for assignment of IP addresses and adjudication of disputes related to IP address assignments.
- The boundaries chosen for class A, B and C addresses were 'unfortunate' in some ways. A class C address might provide for too few computers in some cases, and too many in others. Class A and class B addresses were relatively few in number and demand eventually exceeded supply.
- To deal with such problems, schemes involving dividing IP addresses at arbitrary bit boundaries were employed.
- These schemes are variously called subnet addressing or classless addressing.
- This generalized A, B, and C addresses with their 8-, 16-, and 24-bit prefixes. Now there could also be, for example, 17-, 18-, 19-, 20-, 21-, 22- and 23-bit prefixes.
- This allowed assigned address ranges to more closely match the needs of the networks.
- In the original addressing scheme, software could look at the first few bits of an address and determine its class and the location of the boundary between the prefix and the suffix.
- For example, if the first three bits of a classful address are 110, it is a class C address and therefore the first 24 bits are the prefix and the last 8 are the suffix.
- When the classless addressing scheme was adopted, it had the effect of chopping up class A, B and C addresses in new ways - so that one could no longer tell where the prefix/suffix boundary was by examining the first few bits of the address.
- Often now, an address mask is used to indicate the location of the prefix/suffix boundary.
- The example in the book can be used to illustrate:
D = 1000 0000 0000 1010 0000 0010 0000 0011 (destination address) M = 1111 1111 1111 1111 0000 0000 0000 0000 (address mask) N = 1000 0000 0000 1010 0000 0000 0000 0000 (network address corresponding to D) N equals (D&M)The boundary between the 1's and 0's in the mask can be used to signify that the prefix/suffix boundary in the destination address is that same location. - Certain 'bit operations' are easy for computers to perform. Routers use address masks and bit operations to efficiently test destination addresses to see how they need to be routed.
- When a packet arrives at a router with destination address D, the router looks in its forwarding table for an entry that 'matches' D. Entries in the routing table contain a network address N and a mask M. The address D 'matches' the entry if N equals (D&M). Other information in a matching table entry tells where to forward the matching packet.
- In the CIDR notation, the prefix/suffix boundary may be expressed just with a slash and number added to the IP address.
- For example: 192.5.48.69/26 means that the IP address is 192.5.48.69 and the prefix has 26 bits.
- If an ISP has this address block available:
128.211.0.0/16 Network Address: 1000 0000 1101 0011 0000 0000 0000 0000 Network Mask: 1111 1111 1111 1111 0000 0000 0000 0000
And if one customer needs twelve IP addresses, and another customer needs nine, then the ISP can make these two assignments:Customer #1 128.211.0.16/28 Network Address: 1000 0000 1101 0011 0000 0000 0001 0000 Network Mask: 1111 1111 1111 1111 1111 1111 1111 0000 Customer #2 128.211.0.32/28 Network Address: 1000 0000 1101 0011 0000 0000 0010 0000 Network Mask: 1111 1111 1111 1111 1111 1111 1111 0000
- This assignment gives both customers the possibility of fourteen usable host IP addresses. (The broadcast and network addresses are forbidden as host addresses.)
- It also leaves many unallocated addresses in the original block that can be assigned to other customers later.
- In the example above of
Customer #1 128.211.0.16/28 Network Address: 1000 0000 1101 0011 0000 0000 0001 0000 Network Mask: 1111 1111 1111 1111 1111 1111 1111 0000
The fourteen allowed host addresses are:1000 0000 1101 0011 0000 0000 0001 0001 1000 0000 1101 0011 0000 0000 0001 0010 1000 0000 1101 0011 0000 0000 0001 0011 1000 0000 1101 0011 0000 0000 0001 0100 1000 0000 1101 0011 0000 0000 0001 0101 1000 0000 1101 0011 0000 0000 0001 0110 1000 0000 1101 0011 0000 0000 0001 0111 1000 0000 1101 0011 0000 0000 0001 1000 1000 0000 1101 0011 0000 0000 0001 1001 1000 0000 1101 0011 0000 0000 0001 1010 1000 0000 1101 0011 0000 0000 0001 1011 1000 0000 1101 0011 0000 0000 0001 1100 1000 0000 1101 0011 0000 0000 0001 1101 1000 0000 1101 0011 0000 0000 0001 1110
- The dotted decimal values of these host addresses range from 128.211.0.17 to 128.211.0.30, and the dotted decimal version of the net mask is 255.255.255.240. These are more difficult for the average person to work out, compared to the case for class-full addresses.
- For example The host addresses allowed for csustan.edu range from 130.17.0.1 to 130.17.255.254, and the mask is just 255.255.0.0.
- Some IP addresses are reserved - they can't be assigned to
hosts because they are used for other purposes:
- 21.15.1 IPv4 Network Address: The result of adjoining a suffix of "all zeros" to a network prefix. It denotes the network itself. Example: 130.17.0.0 denotes the csustan.edu network. Another example is the address 128.211.0.16/28 we saw in sections 21.13 and 21.14 of the text.
- 21.15.2 IPv4 Directed Broadcast Address: The result of adjoining a suffix of "all ones" to a network prefix. This address can be used from outside the network to send a copy of a packet to every host on the network. Example: 130.17.255.255
- 21.15.3 IPv4 Limited Broadcast Address This is the IP address 255.255.255.255 - all bits are 1's. This address is used by a host to send a packet to every computer on the directly-connected network. A booting system that has not yet learned anything about the network to which it is connected may use the limited broadcast address to send out a request for information.
- 21.15.4 IPv4's This Computer Address: There is an IP protocol a booting computer can use to broadcast a request that asks: "Will somebody please tell me my IP address?" The sending computer has to put something in the packet field for the source address. The IP address 0.0.0.0 is used for this purpose. We might say that 0.0.0.0 means "me".
- 21.15.5 IPv4 Loopback Address: Any IP address of the form 127.x.y.z is a loopback address. A loopback address is used to test networking software. Packets sent out the loopback interface go down the protocol stack and back up again, but are not transmitted onto the network.
- None of these addresses is to be assigned as a host address.
- Such addresses are to be used only for the indicated purposes.
- Many years ago, developers of early TCP/IP software made an error, causing the protocol software to use network addresses as if they were directed broadcast addresses. As a result, some compatibility problems continue to this day.
- Although we often think of an IP address as being assigned to a computer, in reality an IP addresss is assigned to a connection between a computer and a network.
- Since an Internet router has connections to at least two different networks, it needs at least two IP addresses.
- There's no requirement that the host parts of a router's IP addresses match. It's OK if they are the same, but they don't have to be. Usually, it is convenient for humans if the host parts of all router addresses are the same.
- Just as a router needs multiple IP addresses, so does a multi-homed host.
- IPv6 encourages multihoming.
- IPv6 allows a network to use more than one network prefix.
- If the ISP for a network changes, then the network prefix will probably have to change, and IPv6 is designed to make that simple, so IPv6 provides for more than one network prefix to be used, for example while in the process of phasing out one network prefix and phasing in another.
- However, IPv6 still does not support automatic renumbering adequately, which would allow applications to cut over automatically to a new network prefix when relaunching.
- IPv6 uses network prefixes and host suffixes in a manner similar to IPv4.
- As with CIDR in IPv4, the bit boundary between prefix and suffix is arbitrary.
- Unlike IPv4 addresses which are two-level hierarchical, IPv6 addresses are three-level hierarchical.
- The IPv6 network prefix starts with a Global Prefix which is used for routing, and which basically stands for the organization within which the network resides.
- The second part of the network prefix is a Subnet number, the idea of which is to identify the particular network within the organization.
- The combination of Global Prefix and Subnet always form the first 64 bits of an IPv6 address. The boundary between the two is variable.
- The host address part of an IPv6 address is always the last 64 bits.
- IPv6 provides addresses with limited scope - for example to be used only on a single network, or within a single organization.
- IPv6 does not have broadcast addresses, but can use multicast addresses instead.
- Every IPv6 address is one of three types: unicast, multicast, and anycast (see figure 21.10)
- Dotted decimal notation is unwieldy for 128-bit addresses, for example
105.220.136.100.255.255.255.255.0.0.18.128.140.10.255.255
- Here is a chart showing different ways to express the parts of such
an address:
Binary Decimal Hex 0110 1011 (105) 69 1101 1100 (220) DC 1000 1000 (136) 88 0110 0100 (100) 64 1111 1111 (255) FF 1111 1111 (255) FF 1111 1111 (255) FF 1111 1111 (255) FF 0000 0000 (0) 00 0000 0000 (0) 00 0001 0010 (18) 12 1000 0000 (128) 80 1000 1100 (140) 8C 0000 1010 (10) 0A 1111 1111 (255) FF 1111 1111 (255) FF
- Instead of dotted decimal, a notation called
colon hexadecimal (colon hex)
is used, in which each 16 bits is expressed as four hexadecimal
digits, for example
69DC:8864:FFFF:FFFF:0:1280:8C0A:FFFF
- There is also a convention called Zero Compression that often
further compacts the representation of an address, for example,
FF0C:0:0:0:0:0:0:B1
becomesFF0C::B1
- One aspect of the importance of zero compression is that 32-bit IPv4
addresses will be mapped to IPv6 addresses that begin with 80 0-bits,
followed by 16 1-bits, and then jthe
final 32 bits identical to the IPv4 address. For example,
the csustan.edu network address
130.17.0.0
which, in binary, is:1000 0010 0001 0001 0000 0000 0000 0000
becomes this 128-bit sequence:Binary Hex 0000 0000 0000 0000 (0) 0000 0000 0000 0000 (0) 0000 0000 0000 0000 (0) 0000 0000 0000 0000 (0) 0000 0000 0000 0000 (0) 1111 1111 1111 1111 (FFFF) 1000 0010 0001 0001 (8211) 0000 0000 0000 0000 (0)
If we write the address above in full colon hex notation, we get:0:0:0:0:0:FFFF:8211:0
and when we use zero compression on that, it reduces to:::FFFF:8211:0
which is a fairly compact and understandable designation.